OBJECT RECOGNITION BY SEGMENTED REGIONS USING MULTIPLE CUES ON OUTDOOR ENVIRONMENT

Abstract

This work describes a method to allow objects for autonomous robot navigation on outdoor environment. The proposition of the method segments and recognizes the object from an image taken by moving robot on outdoor environment. We classify the object natural and artificial. We classify trees as natural objects and buildings as artificial objects. Then we define their characteristics individually. In the process, we segment objects included by preprocessing using multiple cues and show the method of segmentation based on low-level features using multiple cues. Multiple cues are color, line segment, context information, HCM (Hue Co-occurrence Matrix), PCs (Principal Components), vanishing point. Objects can be recognized when we combine predefined multiple cues. The correct object recognition of proposed system is over 92% among our test database which consist of about 1200 images. We accomplish the result of image segmentation using multiple cues and object recognition through experiments.